Sep 16, 2023Liked by Alberto Romero

Very interesting article, Alberto, thanks for sharing!

I find the comparison between watching humans at a competitive sport and creative arts doesn't resonate with me. I too enjoy much more watching humans play chess than AIs for all the reasons you mentioned in your article. But taking that argument and transferring it to creative arts like painting, composing, and writing makes little sense to me.

Chess has very clear rules and objectives that allow direct comparison between humans and I enjoy it very much to watch humans compete against each other. But creative arts, for me, is a different story. I don't view it as a direct competition between artists (humans, AI, or mixed) that "holds my breath" like I do with competitive sport like chess. And that's the crucial difference for me

In my opinion, your article conflates liking the creator with liking the creation. Appreciating a great novel or painting is about the work itself, not the identity of the creator (at least for me). If AI can produce writing and art just as original, meaningful, and emotionally resonant as humans, I might like it just as much as if it were created by humans.

Expand full comment

Thanks for the comment Heiko, very important caveat there. I thought about this as I was writing the piece. Let me try to respond.

I agree that artistry doesn't compare with sports in that it's not competitive, but my point is not competition per se. I don't only like to watch Magnus play because he competes against another person (or AI) but also because he makes chess beautiful to watch while I can still identify with him as a fellow human.

I understand competition matters a lot to some people, but my argument is really about something else. I used chess because it's the best example we have where AI has mastered creativity in a way that resembles humans yet we keep watching humans play. I agree part of that is the inherent competitive factor but what I argue is precisely that it's not the only factor at play - that the fact that it's humans doing the thing matters *a lot* to us.

You say that the article "conflates liking the creator with liking the creation" and I say those two things are inevitably entwined. I like Magnus' chess because he's Magnus or I like Magnus because of his chess expertise? It's both at the same time. This happens all the time with art and writing and those other, non-competitive forms of creativity.

We can appreciate a great painting by itself but you can't just ignore authorship, the origin, its history, etc. Those more abstract, less visible aspects of creativity are always there, unseen but present.

And to the last point, which is what I thought didn't quite work with my argument. You say, "If AI can produce writing and art just as original, meaningful, and emotionally resonant as humans, I might like it just as much as if it were created by humans." And this is a great point because I have to agree with you. Those people who saw the cover image of this article thought it was great until they realized it was AI-made. They liked it. And, counterintuitively, this behavior makes my point for me: only when we're "tricked" into thinking something is human-made when it's in fact AI-made we can turn off the fact that what we like is humans doing stuff. We don't reject (some people, partly) AI creations because they are creative but because of the missing human component. The more human presence in the creation, the more we like it *if we know it*.

My argument isn't that we won't like it when AI systems do it - people liked AlphaZero games - but that the fact that they do it better than us matters *very little* because we humans don't just seek perfection or the best of the best, but also familiarity. We like humans and it's impossible for any non-human to compete with that, even if they can compete with the creation, they can't with the creators. That idea that we can isolate creation from creator is quite common, but I don't think it's true. It's not that I'm conflating them but that it's impossible to dissociate one from the other.

As a final argument for why I think being human matters, just think about why do we need to anthropomorphize AI systems. Why do we have that urge to make them feel, seem, and appear more human than they are?

Expand full comment

Alberto writes…. “that the fact that it's humans doing the thing matters a lot to us.”

Yes, it matters to US a lot NOW, because we are cultural primitives living at the dawn of the AI age, so we are obsessed with the human vs. AI comparison.

Most of the AI commentary is focused on AI as it is now, and humans as we are now. Ok, fair enough, but the status quo of now is not going to last, like all “nows” it’s a temporary phase.

I don’t think most readers really grasp how far along we are in this transition already.

A great many of us have already largely traded real life face to face human relationships for digital strangers on the Internet. Why have we done that? Because we have more control over our experience when interacting with digital humans, it’s easier and more convenient to get what we want. Alberto will talk about AI all day long, but my next door neighbor will not.

Where we are now seems like a half way point to where we are headed.

1) real humans => real humans

2) real humans => digital humans

3) real humans => AI entities

The same dynamic which has me talking to you instead of my next door neighbor will continue to develop.

At some point I will trade you in, and replace you with AI, because damnit Alberto, you don’t look at all like Diane Lane. :-) You aren’t available instantly 24/7 at my command. You don’t always write about exactly what I wish to read, when I wish to read it. And so on…

What I won’t attempt to predict is how fast this transition will unfold. As best I can tell, not even the industry experts can speak confidently to that, so I’m not going to pretend that I can.

Expand full comment

Great post, Alberto.

I think about this dumb game I play all the time: Two Dots. I just make squares and stuff, and I'm incredibly cognizant that an AI could solve this puzzle 100x faster (and much more efficiently) than me. I also really enjoy crossword puzzles, which I have not competitive business with against an AI.

Similarly, I do Brazilian Jiu Jitsu consistently not because I think I can defend myself against someone with a gun, or even because I'm worried about defending myself at all, but because it's hard.

We think the hard things matter.

Expand full comment

Thanks Andrew. I agree with you. The fear that comes from the idea that we need to be the best of the best is weird sometimes!

Expand full comment

So funny, when you wrote “humans like imperfection”, I thought, no, humans just like to watch humans.

And then you wrote those exact words. Humans like humans. That’s it. I love your conclusion and think you hit the nail right on the head with this piece. Keep ‘em coming!

Expand full comment

Thank you Jurgen!

Expand full comment

Great essay. What you describe is actually humanity's superpower. Social learning. There's a heart and soul AI doesn't have.


Expand full comment

Thanks Michael. I like that, humanity's superpower

Expand full comment

I've written a few essays on that specific topic. Another one is AI Is[n't] Killimg Artists.

Expand full comment

Your essay got me thinking about how newer camera lenses often intentionally re-introduce imperfections and aberrations to regain some of the charm and unpredictability lost with technical excellence. Just as we prefer a human-played chess match over AI, we may favor a flawed vintage lens over a clinically perfect modern one.

So even in domains like manufacturing, there are parallels to be drawn with art and creativity. Imperfections made by humans are often cherished, not just tolerated.

However, as AI starts replicating more objects and processes, perhaps it too could be tuned to purposefully inject quirks and randomness found in human work.

You've highlighted a fascinating human preference for our fellow flawed creators. While AI may someday mimic this unpredictability, it's an open question whether it can reproduce the essence we appreciate in objects and art produced by human hands and minds.

Expand full comment

Right! I think about this with LoFi music. We don't always like perfection. I find it's tangentially related to how humanity has kind of unlocked the secret sauce of the culture of entertainment and we're just repeating and repeating the formula. But it's becoming too perfect and losing the original appeal - I see the same happening with the creations of AI and machines. I'm sure you know about this much more than me, I find it fascinating and have seen examples in cinema, music, art...

Expand full comment
Sep 16, 2023Liked by Alberto Romero

Yup, this is exactly why no matter how far AI advances, we never have to fear it replacing these imperfections that make us uniquely human.

Expand full comment

Great points. We always come out on top of our greatest fears and challenges – that pushes our creativity more than anything else.

Expand full comment
Sep 15, 2023Liked by Alberto Romero

Thank you! We need to be reminded that we will still be needed even when/if we reach AGI that trumps human intelligence in all areas. We need posts like this.

Expand full comment

We need it, agreed. That doesn't make it true.

What's true is that as soon as our corporate overlords can figure out how to replace us with machines that can do the job cheaper, we will be discarded.

Expand full comment
Sep 17, 2023Liked by Alberto Romero

Sure, human labor will be replaced in companies when it comes to creating products or providing services. But the point of this post as I understand it is that when it comes to recreation, there will always be people who want to see the creations and accomplishments of other human beings (even if AI would do it better according to every single performance metric), and in that sense, we are unreplaceable.

Expand full comment

Agreed, there will always be people who prefer to see content created by humans. But vast swaths of the population don't care about quality writing, or quality anything. So while there will always be a market for human writing, AI content will reduce that market substantially, maybe radically.

In the past, and largely still today, the reading public has no where to go but to their fellow humans for content. We human content creators get the entire market, because there is no other option.

AI will a be strong competitor that will siphon off some large percent of that market, leaving human writers with fewer readers, and thus lower business prospects.

Just as in other industries, there are compelling business reasons to automate content creation where ever possible. As example, if Substack had an AI which could generate content people are willing to pay for, they could increase their cut of the generated income from 10% to 100%, making them radically more profitable.

If they could get rid of most of the writers, they could get rid of many of the employees too, resulting in dramatic cost savings, and a richer bottom line.

Expand full comment

I get your point Phil, but there's one assumption here that breaks the premise: Substack doesn't prefer AI writers, it prefers human writers. And I don't mean just Substack readers, but Substack as a company.

That's the key, they, as humans, like to read what humans have to write. Only deception (passing AI-made as human-made) can bypass that. And we should treat that as an exception, not the norm.

There will be places (other companies, which focus on content rather than humanness) that will gladly take 100% instead of 10%. Sure. No one of us would be there - because we like humans.

Expand full comment

Hi Alberto, good come back, thanks. Two replies...

1) At some point, we're not going to be able tell the difference between AI content and human content. Vast swaths of the population probably already can't tell the difference. I'm not sure I can. More to the point, vast swaths of the population don't care about the difference, just as they don't care that some of the characters in movies are CGI generated. If we can't tell the difference, what impact does a preference for human content have? How would we express such a preference in that situation?

2) My best guess understanding (which could be wrong) is that Substack is not yet profitable. If that's true, then they are burning through cash on hand, and investors will only keep faith for so long. Point being, companies don't always have a choice about such things. If they want to survive they have to compete in the marketplace that exists, not the marketplace they might wish exists.

If other companies can sell mass produced AI content, and keep 100% of income, they would presumably have more financial muscle than those content providers who decline to compete based on some outdated ideological preference.

I largely agree with you in regards to those of us born in to the world before AI. We're going to have lots of opinions about AI vs. human.

The turning point may be with those coming generations born in to a world where AI is everywhere. They may see the AI vs. human equation that obsesses us as being quaint, something ignorant old people are concerned with.

Expand full comment

Estoy de acuerdo con este boletín. Sin embargo, terminas con esta frase:

 "La IA no puede hacer que los humanos se vuelvan obsoletos porque a los humanos les gustan demasiado los humanos.

Por definición, la IA no puede competir con eso y nunca lo hará."

Pienso que para este siglo aplica muy bien, a los humanos nos gustan los humanos. Pero probablemente en el siglo 22, tal vez, la IA y los humanos seamos uno solo. Implantes de brazos, piernas, robóticos. Nanotecnología, ojos robotizados, órganos impresos con inteligencia, implantes de chip cerebrales. 

Sé que estoy hablando desde la ciencia ficción, sé que estoy hablando de prototipos que aún no se han utilizado en humanos o que no ha sido posible su compatibilidad, pero existen estas posibilidades.  

Para mí la IA ganará cuando se empiece a integrar a nosotros los humanos, cuando empiece a modelar nuestras vidas, cuando cada decisión lo hacemos basada en sus recomendaciones. Cuando un humano con inteligencia artificial integrada juegue ajedrez o pinten un cuadro. Por eso pienso que el boletín aplica perfecto para los humanos de esta época. 

Probablemente en el siglo 22  empecemos hablar de otro tipo de seres humanos donde no cuestionamos la IA, porque la IA hace hará parte de esa especie de humanos, será como "órgano" más.


Expand full comment

The chess analogy is a great one, and I agree to a degree.

I think the continuation of the analogy, though, is... what if Magnus Carlson could use Stockfish or another engine while competing. And what if he didn’t have to tell you which was his move and which was inspired by the computer? And what if it was legal for this to happen?

The same goes for art. Does it matter if the artist used AI for 99% of the work or inspiration? How about 50%? What’s the line?

Expand full comment

Well articulated points per the recent state of affairs.

Reframing creativity going forward, as tech and human behavior continue to evolve, one might want to consider:

1. Qualitative leaps in AI capabilities are unfolding at an accelerating rate.

It is extremely unlikely that predictions will prove true in the realm of complex unknown even within months, not to mention years and decades to come. Impact on what constitutes “creativity” may unfold in ways that are yet to be imagined…

2. Entire paradigms and knowledge as we know it are shifting.

We have just begun this epochal transformation.

The very conceptualization used in this article, defining creativity, art, culture, civilization, design and what is human is nonlinear, changing at unprecedented velocity and scale.

3. The capabilities of AI are already groundbreaking, and too early to define, as it encompasses many dimensions that result in emergence as interactions occur btw creative humans and other machines, e.g. real time, vast info processing, best of the best cherry picking and then reshaping (with some deliberate imperfections,,,), human manipulating, desire pleasing, and statistical reshuffling with some underlying rules (extending/redesigning and inventing/imagining).

4. While human creativity is inherent to evolution, epistemology, design and all conscious endeavors,

it is not the sole benchmark for various types of creation, such as 'being human', 'artistic creativity', 'behavioral ingenuity', 'impactful novelty', etc. What we attribute to creativity is currently limited to what we know, have experiences, and acknowledged socially.

New patterns and utterly new levels of creativity are not only possible, but likely and now with new found levers also transpiring across disciplines. A new types of plays are in town, w unprecedented capabilities, some human like, some super human, some alien, some machine, and some yet to emerge, we should perhaps embrace and open up to new found creativmess... here’s to exploration and reimagination beyond(:

Expand full comment

La creatividad es innata al ser humano pero es cierto que se la debe alimentar para que se desarrolle y nos haga felices. Todo lo humano nos interesa, es muy cierto, somos una especie cotilla. Hay una pregunta en el aire, ¿al igual que en anteriores revoluciones tecnológicas vamos a permitir los humanos otra vez que una élite fosilizada se apropie para su solo provecho de las ventajas de la IA mientras el resto, el 99% de la humanidad, solo sufre sus consecuencias? Las teorías del decrecimiento y del reparto justo podrían impulsar una utilización de la IA equilibrada, justa y beneficiosa para la mayoria por ejemplo con más tiempo libre, menos tiempo de trabajo, reparto decente de la riqueza, ocupación laboral y profesional digna para todos...

Expand full comment

Hi Alberto!

You write, "I wonder where the fear that AI will make human creativity obsolete comes from."

It might help to clarify this a bit. AI won't make human creativity obsolete. But it will increasingly threaten the human creativity BUSINESS. People will still write and make art, it will just be ever harder to make a living at it.

As example, if Substack could get an AI which could create content that people will pay for, Substack could keep ALL the income generated by this network instead of just 10%, making it radically more profitable. The same business equations that have caused industry to automate everything it can will play out here too, if/when AI can deliver content that people are willing to buy.

Much of this depends on what time scale we're talking about. Most of the AI commentary I've seen is referencing AI as it is today. I would guess the AI we see today might be reasonably compared to the Internet in 1995. We're at the very beginning of the AI revolution, and there is likely much more to come. No one knows what the future will hold, but it seems unlikely today's AI is the end of the road.

One interesting question is the issue of how far AI can improve. Is there some inherent limit to where this technology can go? Or can AI continually improve over coming decades?

There is a great deal of commentary from Substack writers about how humans love humans and we'll never give each other up etc. I read this as being mostly very understandable wishful thinking from a culture unready to face the future.

It's more accurate to state that humans love what other humans can do for them. Where AI can provide the same benefits humans provide each other, it will be a strong competitor.

As example, consider an AI friend. Your AI friend is ready to be with you 24/7/365 at a moment's notice. No human can match that. Your AI friend involves no compromise or negotiation. No human can match that. Your AI friend is happy to provide you with ceaseless validation, one of the key services we seek from each other. Your AI friend can be customized to your specifications, and if you change your mind 10 minutes later, nobody's feelings get hurt, no ego melodrama, no guilt, no breakup etc.

Today's writers will claim such services will be just a niche for lonely people, or senior citizens in nursing homes etc, and that they wouldn't use an AI friend service themselves because blah blah etc etc. And that's true, for those of us born in the world before AI. And it's NOT true for those coming generations who will be born in to world where AI is everywhere, and is thus considered completely normal.

I'm coming around to the theory that the younger a person is, the harder it is for them to grasp the revolutionary nature of the changes that are coming, because they haven't yet had a chance to live through that many changes. When I was a young man none of us gave computers, the Internet, and of course AI, a single thought. Not a thought. These topics NEVER came up in conversation, as they had no relevance to the lives we were living at that time. It wasn't that long ago.

Point being, whatever you young folks think you know about the future, it's most likely more wrong than you can imagine. :-) Just as it was for us way back when.

Expand full comment